One prompt, one report, three very different approaches to summarizing a 200+ page document.
After months of testing local LLMs, I found that productivity depends on tools, not just models.